Introduction to R for Epidemiologists

Transcription

1 Introduction to R for Epidemiologists Jenna Krall, PhD Thursday, January 29, 2015

2 Final project Epidemiological analysis of real data Must include: Summary statistics T-tests or chi-squared tests Regression Figures Can use provided dataset OR you may provide your own If using your own data, it: Must have at least 6 variables (with at least 2 continuous variables) Must have at least 100 observations Must be able to answer a relevant question (e.g. is air pollution associated with mortality?) You must have your data approved by me by March 5

3 Outline 1. Introduction to base plotting 2. Customizing plots 3. Multiple figures 4. Margins 5. Other plots 6. Saving plots 7. Rules for displaying data

4 Introduction to base plotting Base R comes with excellent graphing capabilities Scatterplots Histograms Box plots Graphical devices (how your computer represents graphical objects) available in R pdf postscript png jpeg

5 Introduction to base plotting base R plots: useful for creating quick plots Other graphics packages exist in R Great for "faceting" Great for multiple panels of plots Other graphics packages ggplot2 lattice Written by Hadley Wickham (RStudio) We will cover ggplot2 later in the term Also useful for multiple panels of plots We will not cover in this course

6 Introduction to base plotting Recall Fisher s iris data data(iris) head(iris) ## Sepal.Length Sepal.Width Petal.Length Petal.Width Species ## setosa ## setosa ## setosa ## setosa ## setosa ## setosa head(iris$sepal.length) ## [1]

7 Introduction to base plotting To create a scatterplot, plot(x, y) plot(x = iris$sepal.length, y = iris$sepal.width) iris$sepal.width iris$sepal.length

8 Introduction to base plotting Histograms show the data distribution for a variable The data distribution is the frequency of different values hist(iris$sepal.length) Histogram of iris$sepal.length Frequency iris$sepal.length

9 Introduction to base plotting Bar plots are used to show the relative frequency of different values of a categorial variable barplot(state.x77[, "Illiteracy"]) Alabama Georgia Maine Nevada Ohio Texas

10 Customizing plots What can we change? Add labels Change colors Change plotting symbol Add multiple plots?plot.default?par

11 Customizing plots Adding labels plot(x = iris$sepal.length, y = iris$sepal.width, xlab = "Sepal length", ylab = "Sepal width", main = "Sepal width vs. sepal length for Fisher's Iris data") Sepal width vs. sepal length for Fisher's Iris data Sepal width Sepal length

12 Colors Use the col argument in the plot function to set color plot(x = iris$sepal.length, y = iris$sepal.width, xlab = "Sepal length", ylab = "Sepal width", col = "red", main = "Sepal width vs. sepal length for Fisher's Iris data") Sepal width vs. sepal length for Fisher's Iris data Sepal width Sepal length

13 Colors We can also use numbers to specify colors: plot(x = iris$sepal.length, y = iris$sepal.width, xlab = "Sepal length", ylab = "Sepal width", col = 3, main = "Sepal width vs. sepal length for Fisher's Iris data") Sepal width vs. sepal length for Fisher's Iris data Sepal width Sepal length

14 Colors We can also use hexadecimal notation (hex) for the combination of red, green, and blue to specify colors: plot(x = iris$sepal.length, y = iris$sepal.width, xlab = "Sepal length", ylab = "Sepal width", col = "#FF00FF", main = "Sepal width vs. sepal length for Fisher's Iris data") Sepal width vs. sepal length for Fisher's Iris data Sepal width Sepal length

15 Colors colors() ## [1] "white" "aliceblue" "antiquewhite" ## [4] "antiquewhite1" "antiquewhite2" "antiquewhite3" ## [7] "antiquewhite4" "aquamarine" "aquamarine1" ## [10] "aquamarine2" "aquamarine3" "aquamarine4" ## [13] "azure" "azure1" "azure2" ## [16] "azure3" "azure4" "beige" ## [19] "bisque" "bisque1" "bisque2" ## [22] "bisque3" "bisque4" "black" ## [25] "blanchedalmond" "blue" "blue1" ## [28] "blue2" "blue3" "blue4" ## [31] "blueviolet" "brown" "brown1" ## [34] "brown2" "brown3" "brown4" ## [37] "burlywood" "burlywood1" "burlywood2" ## [40] "burlywood3" "burlywood4" "cadetblue" ## [43] "cadetblue1" "cadetblue2" "cadetblue3" ## [46] "cadetblue4" "chartreuse" "chartreuse1" ## [49] "chartreuse2" "chartreuse3" "chartreuse4" ## [52] "chocolate" "chocolate1" "chocolate2" ## [55] "chocolate3" "chocolate4" "coral" ## [58] "coral1" "coral2" "coral3" ## [61] "coral4" "cornflowerblue" "cornsilk" ## [64] "cornsilk1" "cornsilk2" "cornsilk3" ## [67] "cornsilk4" "cyan" "cyan1"

16 Colors

17 Colors plot(x = iris$sepal.length, y = iris$sepal.width, xlab = "Sepal length", main = "Sepal width vs. sepal length for Fisher's Iris data", ylab = "Sepal width", col = "dodgerblue") Sepal width vs. sepal length for Fisher's Iris data Sepal width Sepal length

18 Colors We can also give col a vector: col_species <- vector(length = length(iris$species)) col_species[iris$species == "setosa"] <- "dodgerblue" col_species[iris$species == "versicolor"] <- "darkorchid" col_species[iris$species == "virginica"] <- "orangered" plot(x = iris$sepal.length, y = iris$sepal.width, xlab = "Sepal length", main = "Sepal width vs. sepal length for Fisher's Iris data", ylab = "Sepal width", col = col_species)

19 Colors We can also give col a vector: Sepal width vs. sepal length for Fisher's Iris data Sepal width Sepal length

20 Plotting symbol plot(x = iris$sepal.length, y = iris$sepal.width, xlab = "Sepal length", ylab = "Sepal width", pch = 2, main = "Sepal width vs. sepal length for Fisher's Iris data") Sepal width vs. sepal length for Fisher's Iris data Sepal width Sepal length

21 Customizing plots Bar plots are used to show the relative frequency of different values of a categorial variable barplot(state.x77[, "Illiteracy"]) Alabama Georgia Maine Nevada Ohio Texas

22 Customizing plots We can use the las argument in the plot function to change the orientation of the axis barplot(state.x77[, "Illiteracy"], xlab = "State", ylab = "Illiteracy rate", main = "Illiteracy rate by state", las = 2) Illiteracy rate by state Illiteracy rate Alabama Alaska Arizona Arkansas California Colorado Connecticut Delaware Florida Georgia Hawaii Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin Wyoming State But now the x-axis labels are outside the plot!

23 Customizing plots Use par to set global plot options head(par()) ## $xlog ## [1] FALSE ## ## $ylog ## [1] FALSE ## ## $adj ## [1] 0.5 ## ## $ann ## [1] TRUE ## ## $ask ## [1] FALSE ## ## $bg ## [1] "transparent"

24 Margins Change margins using par(mar = c(bottom, left, top, right)) Let s first look at the default We select the mar element from par() mar.default <- par()$mar mar.default ## [1] Suppose we want to increase the bottom margin by 2 par(mar = c(7.1, 4.1, 4.1, 2.1)) plot(x = iris$sepal.length, y = iris$sepal.width, xlab = "Sepal length", ylab = "Sepal width", main = "Sepal width vs. sepal length for Fisher's Iris data")

25 Margins Sepal width vs. sepal length for Fisher's Iris data Sepal width Sepal length Plotting area extends here

26 Customizing plots par(mar = mar.default + c(5, 0, 0, 0)) barplot(state.x77[, "Illiteracy"], xlab = "State", ylab = "Illiteracy rate", main = "Illiteracy rate by state", las = 2) Illiteracy rate by state Illiteracy rate Alabama Alaska Arizona Arkansas California Colorado Connecticut Delaware Florida Georgia Hawaii Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin Wyoming State

27 Customizing plots par(mar = mar.default + c(5, 0, 0, 0)) barplot(state.x77[, "Illiteracy"], xlab = "", ylab = "Illiteracy rate", main = "Illiteracy rate by state", las = 2) mtext("state", side = 1, line = 8) Illiteracy rate by state Illiteracy rate Alabama Alaska Arizona Arkansas California Colorado Connecticut Delaware Florida Georgia Hawaii Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin Wyoming State

28 Sizing plot(x = iris$sepal.length, y = iris$sepal.width, xlab = "Sepal length", ylab = "Sepal width", main = "Sepal width vs. sepal length for Fisher's Iris data") Sepal width vs. sepal length for Fisher's Iris data Sepal width Sepal length

29 Sizing Decrease size of plotting points plot(x = iris$sepal.length, y = iris$sepal.width, xlab = "Sepal length", ylab = "Sepal width", cex =.1, main = "Sepal width vs. sepal length for Fisher's Iris data") Sepal width vs. sepal length for Fisher's Iris data Sepal width Sepal length

30 Sizing Increase size of axis labels and axes: plot(x = iris$sepal.length, y = iris$sepal.width, xlab = "Sepal length", ylab = "Sepal width", cex.lab = 1.5, cex.axis = 2, main = "Sepal width vs. sepal length for Fisher's Iris data") Sepal width vs. sepal length for Fisher's Iris data Sepal width Sepal length

31 Sizing Look under cex for?par plot(x = iris$sepal.length, y = iris$sepal.width, xlab = "Sepal length", ylab = "Sepal width", cex.main =.7, main = "Sepal width vs. sepal length for Fisher's Iris data") Sepal width vs. sepal length for Fisher's Iris data Sepal width Sepal length

32 Layering plots What if we want to add a horizontal line for the mean sepal width? plot(x = iris$sepal.length, y = iris$sepal.width) iris$sepal.width iris$sepal.length

33 Layering plots What if we want to add a horizontal line for the mean sepal width? plot(x = iris$sepal.length, y = iris$sepal.width) abline(h = mean(iris$sepal.width), col = "red") iris$sepal.width iris$sepal.length

34 Layering plots plot(x = iris$sepal.length, y = iris$sepal.width) abline(h = mean(iris$sepal.width), col = "red") abline(v = mean(iris$sepal.length), col = "blue") iris$sepal.width iris$sepal.length

35 Layering plots Adding a specific point plot(x = iris$sepal.length, y = iris$sepal.width) points(x = mean(iris$sepal.length), y = mean(iris$sepal.width), col = "red", pch = "+", cex = 2) iris$sepal.width iris$sepal.length

36 Layering plots We can also add loess line to scatterplot Used to assess direction and magnitude associations or specify breakpoints for regression splines plot(x = iris$sepal.length, y = iris$sepal.width) lowess_iris <- lowess(x = iris$sepal.length, y = iris$sepal.width) lines(x = lowess_iris$x, y = lowess_iris$y, col = "red") iris$sepal.width iris$sepal.length

37 Multiple figures Sepal width vs. sepal length for Fisher's Iris data Sepal width vs. petal length for Fisher's Iris data Sepal width Sepal width Sepal length Petal length Petal width vs. sepal length for Fisher's Iris data Petal width vs. petal length for Fisher's Iris data Petal width Petal width Sepal length Petal length

38 Multiple figures The mfrow option in par allows us to plot multiple figures in one plot Takes the form par(mfrow = c(number of rows, number of columns)) # Change par to allow multiple figures par(mfrow = c(2, 2)) # Create four plots plot(x = iris$sepal.length, y = iris$sepal.width, xlab = "Sepal length", ylab = "Sepal width", main = "Sepal width vs. sepal length for Fisher's Iris data") plot(x = iris$petal.length, y = iris$sepal.width, xlab = "Petal length", ylab = "Sepal width", main = "Sepal width vs. petal length for Fisher's Iris data") plot(x = iris$sepal.length, y = iris$petal.width, xlab = "Sepal length", ylab = "Petal width", main = "Petal width vs. sepal length for Fisher's Iris data") plot(x = iris$petal.length, y = iris$petal.width, xlab = "Petal length", ylab = "Petal width", main = "Petal width vs. petal length for Fisher's Iris data")

39 Saving plots Always set your working directory before saving your plots Saving your plot as a png (portable network graphic): Height and width are in pixels (default is 480 by 480) png("iris_scatterplot.png", height = 700, width = 480) plot(x = iris$sepal.length, y = iris$sepal.width, xlab = "Sepal length", ylab = "Sepal width", main = "Sepal width vs. sepal length for Fisher's Iris data") dev.off() ## pdf ## 2

40 Saving plots Saving your plot as a pdf (portable document format): Height and width are in inches (default is 7 by 7) pdf("iris_scatterplot.pdf", height = 11, width = 7) plot(x = iris$sepal.length, y = iris$sepal.width, xlab = "Sepal length", ylab = "Sepal width", main = "Sepal width vs. sepal length for Fisher's Iris data") dev.off() ## pdf ## 2

41 Saving plots What if you don t run dev.off()? Graphics device does not close You will not have your desired output Can create multiple pages of plots pdf("iris_scatterplot_2.pdf") plot(x = iris$sepal.length, y = iris$sepal.width, xlab = "Sepal length", ylab = "Sepal width", main = "Sepal width vs. sepal length for Fisher's Iris data") plot(x = iris$petal.length, y = iris$petal.width, xlab = "Petal length", ylab = "Petal width", main = "Petal width vs. petal length for Fisher's Iris data") dev.off()

42 Last week s plot # Load google flu data load("googleflu.rdata") # Sort by date flu <- flu[order(flu$date), ] # Find years for data year <- substr(flu$date, 1, 4) # Subset to 2013 flu <- flu[year == "2013", ] # Plot time series for 2013 plot(flu$date, flu$atlanta, type = "b", xlab = "Date", ylab = "Google flu activity in Atlanta", main = "Google flu activity in Atlanta in 2013") abline(v = as.date(" "), col = "red") text(labels = "Class 1", x = as.date(" "), y = 1000, col = "red") abline(v = as.date(" "), col = "red") text(labels = "Class 5", x = as.date(" "), y = 1000, col = "red")

43 Last week s plot Google flu activity in Atlanta in 2013 Google flu activity in Atlanta Class 1 Class 5 Jan Mar May Jul Sep Nov Jan Date

44 Rules for displaying data